Molecular Models for Murine Sperm-Egg Binding*

Murine sperm initiate fertilization by binding to the specialized extracellular matrix of mouse eggs, known as the zona pellucida. Over the past decade, powerful genetic, biophysical, and biochemical techniques have been employed to gain new insights into this interaction. Evidence from these studies does not support either of two major models for binding first proposed over two decades ago. Two more recently established models suggest that protein-protein interactions predominate during this initial stage of fertilization. Another model proposes that about 75–80% of the murine sperm bound to zona pellucida under well defined in vitro conditions is carbohydrate dependent, with the remaining sperm bound via protein-protein interactions. Mounting evidence suggests that the carbohydrate sequences coating the murine egg could be employed as specific immune recognition markers. Continued investigation of this system may resolve many of these controversial findings and reveal novel functions for murine zona pellucida glycoproteins.


The Emergence of Major Models for Murine Sperm-Egg Binding
Two major hypothetical models for murine gamete binding were first presented in 1985. The resistance of mZP3 to denaturing conditions suggested that its protein-linked glycans could be responsible for initial sperm-egg binding (10). Subsequent studies demonstrated that either glycopeptides or O-linked glycans derived from mZP3 inhibit murine sperm-egg binding in a competitive murine sperm-egg binding assay (10,11). Therefore Wassarman and colleagues suggested that EBPs on the sperm plasma membrane interact with mZP3 associated O-glycans to mediate initial murine sperm-egg binding.
Digestion of the O-glycans derived from mZP3 with ␣-galactosidase results in the loss of their inhibitory activity in the murine sperm-egg binding assay, suggesting that terminal Gal␣1-3Gal sequence are crucial for binding (12). However, a later study confirmed that both terminal ␤l-4-linked Gal and its ␣1-3-galactosylated analogue (Gal␣1-3Gal␤1-4) are recognized during initial gamete binding (13). Insertion of the mZP3 gene into the F9 embryonal carcinoma cells yields recombinant mZP3 that is functionally equivalent to native mZP3 (14). Subsequent exon swapping studies suggested that a region from Ser-329 3 Ser-334 coded for by exon 7 of the mZP3 gene (SNSSSS) is necessary for inhibiting murine sperm-egg binding in vitro (15). Conversion of either Ser-332 or Ser-334 (SNSSSS) to Ala results in mutant rec-mZP3 that lacks the capacity to inhibit murine sperm-ZP binding (16). Therefore vicinal O-glycans at Ser-332 and Ser-334 were proposed to be essential for binding. A specific EBP designated sp56 identified by cross-linking experiments (17) was hypothesized to interact with these vicinal O-glycans on mZP3 (18).
Another carbohydrate-dependent model suggests that a specific ␤4GalT expressed on the murine sperm head mediates binding by interacting with ␤-linked GlcNAc residues at the terminal ends of O-glycans linked to ZP3 (19). Shur and colleagues (20) further demonstrated that these GlcNAc residues are expressed on O-linked polylactosaminoglycans associated with mZP3. They also suggested that an N-acetylglucosaminidase released during the cortical reaction removes terminal ␤-linked GlcNAc residues on mZP3, thus leading to a specific block to polyspermy (21).

Modern Studies on the Major Models
These two models were completely dominant in the murine system for the decade following their proposal. The development of molecular biological approaches (22) and ultrasensitive mass spectrometric methods for sequencing glycans (23) and proteins (24) enabled investigators to directly test these predominant models.
Female transgenic mice lacking ␣1-3-galactosyltransferase (and thus terminal Gal␣1-3Gal sequences) retain their fertility (25) and produce eggs that display a normal sperm binding phenotype (26), confirming that terminal Gal␣1-3Gal sequences are not required for binding as proposed originally (12). Proteomic analysis of mZP3 indicated that Ser-332 and Ser-334 are not O-glycosylated in native mZP3, consistent with the computer prediction that these Ser residues are not preferred glycosylation sites (24). Other studies confirmed that sp56 is located within the acrosome (27,28), precluding its participation in initial sperm-egg binding before the acrosome reaction.
Wassarman and co-workers have consistently contested these findings (29,30). In a recent study, sp56 was again localized to the sperm plasma membrane (29). This group recently suggested that only those mZP3 molecules that actually coat the surface of the egg are modified at Ser-332 and Ser-334 (30). This restricted modification would be extremely difficult to detect because the molar level of mZP3 on the surface of the mZP is quite low.
The model suggesting that the ␤4GalT acts as the major EBP was also undermined by more recent data. Murine eggs bind three to four more times more mutant sperm lacking ␤4GalT(Ϫ/Ϫ) than sperm from wild type mice (31). Sensitive biophysical analyses have confirmed that O-linked polylactosaminoglycans are not expressed on the mZP (32,33). This matrix marginally expresses terminal ␤-linked GlcNAc sequences on both its N-and O-glycans (32,34), a result consistent with lectin binding studies involving freshly ovulated eggs (34).

Carbohydrate-independent Models for Murine Gamete Binding
The two established models for murine gamete binding have been severely compromised by evidence collected over the last decade. This lack of a consensus has led to the development of alternate molecular models for initial sperm-egg binding.
Shur and colleagues recently proposed that the ␤4GalT "sees" ZP3 oligosaccharides in the context of another component (SED1) that is responsible for the initial docking of sperm to the mZP (35). SED1 is homologous to a small group of secreted cell matrix adhesive proteins that contain Notch-like EGF repeats and discoidin/F5/8 type C domains. This discoidin/C domain mediates binding to the mZP based on studies with the truncated forms of SED1. Matings of male SED1 knock-out mice with wild type females results in a 50 -60% decrease in litter size compared with control matings (35). Rodeheffer and Shur (36) recently reported the existence of a 250-kDa glycoprotein specifically associated with ovulated mouse eggs that mediates mZP3-independent bind-ing of sperm to the mZP. However, there is insufficient protein for proteomic analysis, so its identity is not established.
Powerful molecular biological approaches have been employed to investigate the specificity of murine sperm-egg binding in the context of the human interaction (22). Human sperm do not bind to murine eggs (37). Dean and his colleagues hypothesized that female transgenic mice expressing huZP3 instead of mZP3 ("huZP3 rescue mice") would produce eggs that could bind human sperm (38). Eggs from the huZP3 rescue mice bind murine but not human sperm. These investigators proposed that the supramolecular complex of all huZP glycoproteins is required for human sperm binding.
To test this hypothesis, mice expressing huZP2 (huZP2 rescue mice) and both huZP2 and huZP3 (huZP2/huZP3 rescue mice) were created (39). However, eggs from both mouse lines continue to bind murine but not human sperm. These findings are more consistent with the model suggesting that glycosylation confers the taxon binding specificity of human sperm (40). A recent study suggests that the huZP is composed of four glycoproteins (41). Therefore Dean and his co-workers (42) have recently suggested that a complex of four rather than three huZP glycoproteins may be required for human sperm binding.

A Model for Carbohydrate-dependent Adhesion
In 1983, Yamagata and colleagues (43) reported that murine sperm bind to rabbit erythrocytes but did not characterize this interaction. A later study revealed very robust binding between these cell types, leading to the formation of aggregates ( Fig. 1, panel A). Only acrosome-intact sperm bind, indicating that this interaction is between their plasma membranes (44) (Fig. 1, panel B). However, the molecular basis of this somatic cell adhesion was not determined. Rabbit erythrocytes express very large glycosphingolipids known as polyglycosylceramides (45). The oligosaccharide component of these polyglycosylceramides is composed of between one and seven branches (Gal␣1-3Gal␤1-4GlcNAc␤1-6) linked to a linear polylactosamine backbone. The largest derivative in this family (PG-7) is shown in Fig. 2. This result suggested that sperm-erythrocyte binding could be mediated solely via terminal Gal␣1-3Gal sequences, based on the evidence available at that time (12). However, exhaustive treatment of rabbit erythrocytes with ␣-galactosidase reduced but did not eliminate sperm binding (46), indicating that glycans terminated with ␤1-4-linked Gal could mediate sperm binding to this cell type and by analogy the mZP (46).
Evidence that polyglycosylceramides could mediate this binding by interacting with sperm EBPs was provided in another study. Artificial oligosaccharide Constructs expressing the same terminal branched sequence as the rabbit erythrocyte polyglycosylceramides (Fig. 2, dashed boxes in PG-7, S1, and S4) were tested in the competitive murine sperm-egg binding assay at a final concentration of 4 M (13). Construct S1 blocks murine sperm-egg binding by 70 -75%, whereas Construct S4 is poorly inhibitory (13). Construct S2 is equally as active as Construct S1 in this assay system, indicating that terminal ␣-linked Gal sequences are not obligatory for inhibitory activity. Construct S5, the monovalent derivative of S2, was poorly inhibitory. Construct S3 lacking terminal ␤1-4-linked Gal did not block murine sperm-egg binding. This finding suggested that the branched polylactosamine sequence in Construct S1 mediates sperm binding to rabbit erythrocytes. However, because terminal Gal␣1-3Gal sequences are not essential for murine spermegg binding (26) or fertility (25), the branched polylactosamine sequence in Construct S2 was implicated in initial murine gamete binding.
Data obtained in other laboratories were consistent with this specificity. Digestion of murine eggs with a highly specific ␤1-4 galactosidase reduces murine sperm binding to mZP by 70% (34). Neuraminidase treatment of murine eggs that exposes underlying terminal ␤1-4-linked Gal sequences increases binding by 30 -40% in the same study. However, ␣-galactosidase treatment has little if any effect in this system (34). Small oligosaccharides terminated with either LacNAc or its ␣-galactosylated analogue (Gal␣1-3Gal␤1-4GlcNAc) maximally inhibit sperm egg binding by 74 -78% but at far higher concentrations than Constructs S1 or S2 (Fig. 2) (47).
There was one major problem with this potential specificity, however. Evidence available at that time indicated that linear, but not branched, polylactosamine type sequences are linked to mZP3 glycans (20,48). There was evidence confirming that mZP3 expresses tri-and tetraantennary N-linked oligosaccharides bearing the ␤1-6-linked LacNAc sequence attached to the ␣1-6-linked core mannose residue (48). This ␤1-6-linked LacNAc sequence is associated with inhibitory Construct S2 (Fig. 2) (13), mZP N-glycans (48), and rabbit erythrocytes following exhaustive digestion with ␣-galactosidase. Based on this overlap, the ␤1-6-linked LacNAc sequence was proposed to be the major carbohydrate ligand recognized during murine gamete binding (44).
This sequence is primarily expressed on Core 2 O-glycans, complex type tri-and tetraantennary N-glycans, and branched polylactosamine sequences in mice. Ultrasensitive biophysical analyses confirmed that the majority of the mZP O-glycans are Core 2 type sequences linked almost exclusively to mZP3 (32,33). Complex type tri-and tetraantennary type N-glycans bearing ␤1-6-linked LacNAc sequences are preferentially associated with mZP3 (33,48). Therefore the sperm receptor glycoprotein mZP3 preferentially expresses oligosaccharides bearing ␤1-6-linked LacNAc sequences. All the current evidence indicates that branched polylactosamine type sequences are not expressed on the mZP, but rigorous analysis will be necessary to confirm that this difference exists.
Another distinct possibility was that rabbit erythrocytes could express proteinlinked oligosaccharides that resemble those found on the mZP. However, a recent preliminary analysis indicates that the majority of the N-glycans from this cell type are biantennary complex type sequences with one antennae consisting of a highly branched polylactosamine type chain exactly like those present in polyglycosylceramides from this cell type (49). These data suggest that murine sperm bind to both N-glycans and polyglycosylceramides via the recognition of the terminal branched polylactosamine sequence present in Constructs S1 and S4 (Fig. 2). However, adhesion is likely achieved via multivalent interactions with monovalent branched polylactosamine sequences expressed on this cell type, rather than with divalent presentations like those involving Construct S1. Constructs S1 and S2 have never been found in the natural setting.

Potential for Redundant Carbohydrate-Protein and Protein-Protein Interactions during Murine Gamete Binding
The binding of murine sperm to mouse eggs is blocked by the exoglycosidase digestion of the mZP (34) or the presence of specific carbohydrate inhibitors (13,47). However, the maximal inhibitory effect obtained in these studies ranges from 70 -78% (13,34,47). Mild periodate oxidation of human ZP glycans also reduces human sperm binding by 79% in the human hemizona assay system (50). These results suggest that perhaps 20 -25% of the sperm binding sites on mZP in these assay systems are carbohydrate-independent.
This redundant binding is supported by the inactivation of glycosyltransferase genes. The obligatory enzyme required for the synthesis of Core 2 O-glycans is the ␤1-6 N-acetylglucosaminyltransferase (Core 2 enzyme). Inactivation of the ovary-specific isoform of this enzyme does not affect the fertility of the resulting mutant mice (51). Deletion of the N-acetylglucosaminyltransferase that adds ␤1-6-linked GlcNAc to the trisaccharide core of N-glycans (Mgat5) generates mutant mice that are deficient in tri-and tetraantennary N-glycans bearing ␤1-6-linked LacNAc sequences (52). However, these mice display only a 40 -50% loss in fecundity. 4 Inactivation of N-acetylglucosaminyltransferase I (Mgat1) in oocytes using the ZP3Cre recombinase transgene system results in mZP glycoproteins that completely lack complex and hybrid type N-glycans, yet are fertilized, yield embryos that implant, and generate heterozygotes that develop to birth (53). These results suggest that the expression of N-or O-glycans bearing ␤1-6-linked LacNAc sequences is not obligatory for fertility. Another essential consideration is that the threshold of residual sperm binding that must be achieved to ensure fertility has not been firmly established. Therefore 20 -25% residual binding may be sufficient to completely recover fertility. However, eggs from these knock-out mice should display deficient sperm-egg binding in vitro compared with wild type eggs.
The molecular basis underlying this carbohydrate independent interaction is currently unknown. The initial events of murine fertilization involve both sperm binding and the induction of the acrosome reaction. The binding of murine sperm to rabbit erythrocytes clearly does not induce the acrosome reaction (44), suggesting that the signal transduction event that triggers this reaction is not carbohydrate-mediated. Thus the binding of a currently unidentified signal transduction molecule or complex on murine sperm to mZP3 may be essential for inducing the acrosome reaction. This interaction could also be responsible for this residual carbohydrate-independent murine spermegg binding, although that hypothesis has yet to be rigorously tested.

Why Continue to Employ Carbohydrate Recognition for Murine Gamete Binding?
It is very easy to see how polar oligosaccharides with very complex sequences could act as very useful "species recognition markers" to ensure rapid and highly species-specific binding in an aquatic environment (reviewed in Refs. 1-3). However, mammals employ rather sophisticated physical stimuli and behaviors to ensure that mating occurs only between members of the same species. Gene deletion studies indicate that murine sperm continue to bind to mouse eggs in the absence of putative carbohydrate ligands. Thus two major questions arise: (i) why would mice continue to employ carbohydrate dependent gamete binding in the uterus; and (ii) why would initial murine sperm-egg binding need to be species-specific under such circumstances?
There exists abundant evidence that glycosylation is essential for many housekeeping functions such as protein folding, stabilization against proteolysis, and specific targeting to intracellular and extracellular compartments (reviewed in Ref. 54). However, in some cases only glycosylation site occupancy is required to fulfill such housekeeping functions. It is only logical that further modification of oligosaccharides at such sites could expand the biological roles of a specific glycoprotein. These added functionalities would be positively selected and could conceivably explain the enormous structural diversity of the oligosaccharides linked to glycoproteins like mZP3. This type of operating logic underlies all studies in the area now known as "functional glycomics." Such carbohydrate-mediated activities would not be immediately obvious from protein structure but could only be exposed by precise carbohydrate sequencing methods in combination with functional analyses.
Therefore the existing precedent in lower species combined with the selection pressure to optimize for sperm binding could explain why carbohydrate recognition remains a major component of mammalian sperm-egg binding. The analysis of the protein-linked mZP glycans has recently revealed another possible function associated with mZP glycans. Eggs from any female member of a mammalian species must be highly protected from potential immune responses to ensure the transmission of that individual's genetic legacy. Murine eggs lack histocompatibility markers for "self" and thus can avoid certain types of deleterious cell mediated responses (reviewed in Ref. 55). However, the egg may require recognition signals for other immune effector cells, especially natural killer cells that target cells lacking histocompatibility markers, a concept known as "missing self" (56). The murine egg could also become susceptible to immune destruction after fertilization when it begins to express foreign paternal histocompatibility markers.
It is therefore quite significant that the majority of the carbohydrate sequences expressed on the mZP are also highly up-regulated on murine cytotoxic T lymphocytes and T helper cells during their activation (reviewed in Ref. 55). These carbohydrate sequences could be employed in the immune system as protective markers to prevent inadvertent destruction of lymphocytes during their activation. It is only logical that eggs would express such sequences to provide "fail-safe" protection against unintended cytolysis. These egg glycans as a set could therefore be considered markers for the immune recognition of "species" rather than self (reviewed in Ref. 55). Species recognition very likely preceded the recognition of self during the evolution of the metazoan immune system. Therefore carbohydrate recognition markers could have been easily integrated into the matrix coating the eggs of the earliest sexually reproducing metazoans to mediate recognition by immune effector cells. This type of "handshaking" could be one of the key components that protects histoincompatible mammalian blastocysts prior to their hatching. There is evidence for this type of system on the murine egg. The ␤1-6linked LacNAc sequence attached to the core ␣1-6-linked mannose residue of tri-and tetraantennary N-glycans is often extended with linear polylactosamine chains (52). Linear polylactosamine type chains are associated with mZP3 based on carbohydrate sequencing data (33,48). Oligosaccharides of this type are thought to interact with galectins on the surface of T cells to drive the formation of lattices that inhibit T cell receptor signal transduction (52). The type of protection may be essential for other cells and tissues, because Mgat5 knock-out mice display not only compromised fertility but very significant autoimmune disorders and a hyperactivated immune response (52).
There is also evidence for this type of protective effect in the human system. Based on lectin binding studies, human sperm, huZP, and egg cell are coated with specific carbohydrate sequences (bisecting type N-glycans) that have been implicated in the suppression of natural killer cells (57). As noted earlier, germ cells should be susceptible to this immune cell type based on the missing self model (56). Recent data suggest that bisecting type glycans are also expressed on persistent pathogens like HIV-1 and aggressive tumor cells (reviewed in Ref. 55). The coupling of gamete protection to the survival of persistent pathogens and tumor cells makes considerable pathophysiological sense. Thus the carbohydrate sequences that coat any metazoan egg need to be carefully investigated for their potential immunomodulatory effects in both natural and pathological states.
As for the issue of species-specific binding, Bedford (37) reported that murine sperm bind to eggs from many different mammalian species. These results are consistent with the concept that no selection pressure for species-specific binding exists in the murine reproductive system. In contrast to the avid binding to heterologous eggs shown by the epididymal sperm of mice and some other mammals, human sperm do not bind to foreign eggs except those of some other hominoids like the gibbon (37) and the lowland gorilla (58). The selection pressure that created this taxon-specific binding of human sperm is currently unknown.
Do these findings indicate that xenogenic sperm recognize exactly the same carbohydrate sequence as murine sperm do during their binding to the murine egg? The answer to this question is not necessarily. Xenogenic sperm could recognize and bind to a specific carbohydrate sequence presented on the mZP that is not employed for murine gamete binding. Such sperm could also recognize the same carbohydrate sequence as murine eggs but as a non-physiological ligand (i.e. not expressed on the ZP of eggs from that xenogenic species). There is data indicating that the binding of murine sperm to rabbit erythrocytes involves a nonphysiological ligand. Another possibility is that xenogenic sperm could bind to murine eggs in whole or in part via protein-protein interactions. Thus the very strong temptation to draw any conclusions about the binding specificity of spermegg interactions from such cross-species binding studies clearly should be avoided. Only a combination of structural and functional analyses has the potential to unambiguously define this precise specificity in any mammalian species.

Conclusion
Our understanding of the initial murine sperm-egg binding interaction has progressed substantially over the past decade, due to the development of powerful genetic, biophysical, and biochemical tools. This area of investigation remains controversial because of the conflicting results and multiple models. As outlined here, two models propose that protein-protein interactions predominate. Another model suggests that murine gamete binding involves both protein-carbohydrate and protein-protein interactions. This latter model is attractive because of the redundancy it offers and the evidence obtained from several laboratories that support it. Continued investigation of the murine sperm-egg binding interaction will eventually lead to a level of understanding that will be useful for practical application in many mammals including humans. This information may also provide insights into other immunological and pathophysiological processes that are only now being appreciated.